基于遗传算法的自适应图像分割技术研究  

Research on Adaptive Image Segmentation Technique Based on Genetic Algorithm

作　　者：李晓芳[1] 尹福成[2] LI Xiaofang;YIN Fucheng(College of Engineering & Technical,Chengdu University of Technology,Leshan 614000;Neijiang Normal University,Neijiang 641000)

机构地区：[1]成都理工大学工程技术学院,乐山614000 [2]内江师范学院,内江641000

出　　处：《计算机与数字工程》2019年第4期930-932,1000,共4页Computer & Digital Engineering

基　　金：国家自然科学基金面上项目(编号:11375055)资助

摘　　要：在数字人重建中,图像分割是数字人重建的关键要素,常规图像分割方法不但效率低,丢失信息严重,更重要的是分割精度极低。为了解决这些问题,提出了一种基于非局部相似正则化降噪方法的改进并将改进双种群遗传算法,通过分割结果表明论文提出的算法具有较高稳定性,分割效果较精确,而且大幅度降低了遗传算法的计算复杂度。Image segmentation is a key factor in reestablishing digital human.The traditional image segmentation method is inefficient and has severe information loss.What's more,the segmentation accuracy is highly low.In order to solve these problems,an improved 2-population genetic algorithm based on non-local similarity regularization is proposed.The segmentation results show that the algorithm has high stability and accurate segmentation effect,and greatly reduces computation complexity of the genetic algorithm.

关 键 词：数字人 图像分割 遗传算法 非局部相似正则化 

分 类 号：TN911.73[电子电信—通信与信息系统]